Isocitrate Dehydrogenase Mutations Induce BCL-2 Dependence in Acute Myeloid Leukemia through Inhibition of Cytochrome C Oxidase Function

IDH1 and IDH2 are two of the most frequently mutated genes in acute myeloid leukemia at an overall frequency of about 15-20%. The genes encode enzymes in the citric acid cycle that normally catalyze the oxidative decarboxylation of isocitrate, producing α-ketoglutarate (α-KG). The mutant enzymes gain a neomorphic activity that catalyzes the conversion of α-KG to (R)-2-hydroxyglutarate (2-HG). The intracellular concentration of (R)-2-HG is over 100-fold higher in IDH-mutated cells than in wildtype cells. (R)-2-HG has been shown to be a competitive inhibitor of multiple α-KG dependent dioxygenases including TET2, and the Jumonji-C domain containing histone demethylases. These enzymes are thought to be the main targets through which (R)-2-HG exerts its effects on leukemogenesis.

We previously reported that inhibition of the anti-apoptotic BCL-2 protein is synthetic lethal against mutant IDH which we discovered through a large-scale pooled lentiviral RNA interference screen. We confirmed that expression of mutant IDH1 or IDH2 strikingly sensitized AML cells to shRNA-mediated BCL-2 knockdown and pharmacologic BCL-2 inhibition with ABT-199, a highly specific BH3 mimetic. Importantly, we found that primary human AML blasts harboring IDH mutations were significantly more sensitive to ABT-199 than blasts with wildtype IDH ex vivo and in xenograft transplant models. Furthermore, we showed that ABT-199 was able to target the leukemic stem cell compartment in IDH-mutated samples.

Here, we present our work to uncover the synthetic lethal mechanism. An important clue to the mechanism surfaced with the finding that treatment with a cell-permeable precursor of (R)-2-HG sensitized AML cells to BCL-2 inhibition, indicating that the intracellular accumulation of (R)-2-HG found in IDH-mutated cells is sufficient to mediate the phenotype. In addition, we found that (R)-2-HG was able to sensitize isolated mitochondria to ABT-199 with collapse of the mitochondrial transmembrane potential as a surrogate marker for commitment to apoptosis. This finding indicates that 1) the target mediating the synthetic lethal phenotype is localized to the mitochondria, and 2) changes in the epigenome and expression of nuclear-encoded genes are not required for the synthetic lethal phenotype.

To identify the potential mitochondrial molecular target, we focused our analysis on the effect of (R)-2-HG on the enzymatic activity of individual complexes in the electron transport chain (ETC), given that ETC dysfunction can potentially alter the threshold for apoptosis. We found that (R)-2-HG at concentrations found in IDH mutated cells inhibited the in vitro enzymatic activity of complex IV (cytochrome C oxidase (COX)) in a dose-dependent manner, but had no effect on the remaining ETC complexes. This finding has in vivo significance as COX activity in intact IDH-mutated primary AML cells was found to be significantly decreased compared with non-mutated AML cells.

Next, we investigated the possibility that COX inhibition is sufficient to induce BCL-2 dependence. We found that suppression of COX activity with chemical inhibitors or genetically through shRNA-mediated knockdown of a COX subunit (COX-IV) was sufficient to sensitize AML cells to ABT-199. Furthermore, treatment with tigecycline, a FDA-approved antibiotic that has previously been shown to disrupt ETC function through inhibition of mitochondrial translation resulting in decreased expression of mitochondrial-encoded proteins including the catalytic subunits of the COX complex, reproduced the sensitization effect in non-IDH mutated AML cells. Based on the above findings, we propose a mechanistic model in which (R)-2-HG accumulation in IDH mutant cells directly inhibits COX, thereby lowering the mitochondrial threshold for triggering apoptosis upon BCL-2 inhibition.

In summary, we discovered that in addition to the previously described inhibition of α-KG dependent dioxygenases through (R)-2-HG production, IDH mutations also affect mitochondrial bioenergetics. This finding opens up the intriguing possibility that IDH mutations contribute to leukemogenesis not only through epigenetic changes but also through metabolic dysregulation. Lastly, our findings form the rational basis for combining agents that disrupt ETC function such as tigecycline with ABT-199 to target resistant cancer cells and maximize the clinical utility of this promising drug.